Plant metabolomics driven chemical and biological comparison of the root bark of Dictamnus dasycarpus and Dictamnus angustifolius

Abstract

The root bark of Dictamnus dasycarpus (DD) and Dictamnus angustifolius (DA), two commonly used traditional Chinese medicines from the genus Dictamnus (Rutaceae), have been widely used for the treatment of skin ulcer, skin inflammation and other related inflammatory diseases in clinical therapy. The present study compares the chemical and biological differences between DD and DA. An ultra-fast liquid chromatography-tandem mass spectrometry (UFLC-MS/MS)-based plant metabolomics approach has been employed to compare the chemical profiles of DD and DA. 16 chemical markers, including 4 degraded limonoids, 3 limonoids and 9 furoquinoline alkaloids, were identified and the results indicated that there were significant differences in the chemical composition between the two medicinal plants. The methanolic extracts of DA have a relatively higher level of many major furoquinoline alkaloids than that of DD, whereas DD contains more degraded limonoids and limonoids. In addition, the cytotoxic, antimicrobial and antioxidant activities of chemical markers and methanolic extracts of these two medicinal herbs were evaluated. The results presented in this study suggest that DD and DA possess cytotoxic, antimicrobial and antioxidant activities. However, the extracts of DA showed significantly stronger activities than DD in these three bioactivity tests mentioned above. Bioassays also suggested that furoquinoline alkaloids may play a major role in the bioactivities of these two plants and thus could be used as references for quality control.

---

1. Introduction

The genus Dictamnus L. (Rutaceae) is mainly distributed in Europe and North Asia and involves five species, only two of which are found in China.¹ Dictamnus dasycarpus Turcz. (DD), which is distributed widely throughout China, has been used as traditional medicine to treat various diseases such as skin inflammation, eczema, rubella, scabies, acute rheumatoid arthritis, jaundice, cold, and headache.² Dictamnus angustifolius G. Don ex Sweet (DA), which is grown only in the XinJiang province of China, has been locally used as an alternative for DD in the local.³ Phytochemical investigations indicated that furoquinoline alkaloids and limonoids were the major bioactive components of the two species.^4,5 Because these compounds showed potential cytotoxic, antimicrobial, neuroprotective, anti-inflammatory, anti-platelet-aggregation and vascular-relaxing activities,^6–8 they have attracted great interest in the past few years.

The chemical analysis of Dictamnus not only offers information about its therapeutic value, but also differentiates botanical species having morphological similarity. Currently, high performance liquid chromatography (HPLC) coupled with mass spectrometry (HPLC-MS) is being increasingly used as the predominant tool for the qualitative and quantitative analysis of the chemical constituents in natural products.^9–11 Although DD and DA have been used to treat similar diseases for a long time, no research has been conducted to compare their chemical profiles, and slight attention has been paid to investigate biological differences between DD and DA.

In the present study, an ultra-fast liquid chromatography-tandem mass spectrometry (UFLC-MS/MS)-based plant metabolomics approach was first employed to reveal the chemical difference between DD and DA. Furthermore, their cytotoxic, antimicrobial and antioxidant activities, which are closely associated with the therapeutic effects of DD and DA in clinics, were also compared (Fig. 1).

Fig. 1 Schematic of workflows for revealing the chemical and biological differences between DD and DA.

2. Experimental

2.1 Chemicals and materials

The reference standards of dictamnine, isodictamnine, robustine, γ-fagarine, skimmianine, fraxinellone, 5-hydroxy-4,8-dimethoxy furoquinoline, isodictamdiol, dictamdiol, obacunone and limonin were isolated by our laboratory with a purity of more than 98%, and their chemical structures were elucidated by comparing their ¹H and ¹³C NMR data with those reported in previous studies.^5,12–14 HPLC-grade methanol was purchased from Merck (Darmstadt, Germany). Deionized water was prepared using a Milli-Q water purification system (Millipore, Bedford, MA, USA). Analytical grade formic acid and dimethyl sulfoxide (DMSO) were obtained from Nanjing Chemical Reagent (Nanjing, China). A total of 20 samples (including 10 batches of DD and 10 batches of DA) were collected from different localities of the Jiangsu province and Xinjiang Province of China in October, 2010. The voucher specimens, authenticated by Prof. Sheban Pu (China Pharmaceutical University), were deposited in the Department of Natural Medicinal Chemistry, China Pharmaceutical University, Nanjing, China.

2.2 Preparation of stock and working solutions of reference standards

Stock solutions of all the above mentioned 11 reference standards were prepared at 1 mg mL⁻¹ with methanol and DMSO. They were further diluted with methanol and used as working solutions with a concentration of 10 μg mL⁻¹.

2.3 Sample preparation

Each sample powder (0.5 g) was accurately weighted and extracted by ultrasonication with 10 mL methanol for 30 min. The extraction solution was transferred into a 10 mL volumetric flask, which was filled up to its volume with methanol. After centrifuging at 16 000g for 20 min, the supernatant was sealed and stored at 4 °C prior to LC/MS analysis and bioactivity assays. Quality control (QC) samples were prepared by mixing equal volumes of each sample analyzed in this study.

2.4 Instrument and conditions

Chromatographic analysis was performed using a Shimadzu UFLC system (Shimadzu, Japan). A 3 μL aliquot of filtered sample solution was injected into a Shim-pack XR-ODS III column (50 mm × 2.1 mm I.D., 1.6 μm, Shimadzu, Japan) held at 30 °C and the flow rate was 0.3 mL min⁻¹. The gradient elution was 30 min with mobile phase A (0.1% formic acid in water) and mobile phase B (methanol). From 0.1 min to 25 min, mobile phase B increased linearly from 5% to 100% and was kept at 100% for 5 min. At 30 min, mobile phase B was adjusted to 5% for equilibration.

Mass spectrometry detection was performed using a Shimadzu triple quadrupole mass spectrometry (8040, Shimadzu, Japan) equipped with electrospray ionization. Positive ionization mode mass spectra were acquired in a full-scan operation with a scan range of 120–1000 m/z. The temperatures of the desorption line and the heat block were maintained at 250 °C and 400 °C, respectively. The flow rate of the nebulizing gas (N₂) and that of the drying gas (N₂) were set at 3 L min⁻¹ and 15 L min⁻¹, respectively. MS² experiments were performed at different levels of collision energy. All data were acquired using the Shimadzu LCMS solution version 5.53 (Shimadzu, Japan).

2.5 Data pretreatment and multivariate data analysis

Automatic peak detection and mass spectrum deconvolution were performed with the MZ Mine 2 software after converting LC/MS raw data into netCDF files.¹⁵ A three-dimensional matrix consisting of sample information, variables characterized by retention time (t_R) and m/z values, and their corresponding intensities were obtained and exported to an Excel table. These data were handled according to the “80% rule”: only those variables with intensities above zero in at least 80% of a group were retained for subsequent analysis.^16,17 Then, the ion intensity for each detected peak was normalized against the sum of the peak intensities within that sample. After normalization, variables with a relative standard deviation (RSD) higher than 30% in the QC samples were removed. Then, the preprocessed dataset was used to perform unsupervised principal component analysis (PCA) and supervised orthogonal partial least squares discriminant analysis (OPLS-DA). All multivariate analysis (MVAs) and modeling were carried out using the SIMCA-P version 13.0 (Umetrics, Sweden).

2.6 Identification and structure annotation of potential chemical markers

The corresponding variable importance in the projection (VIP) value was calculated in the OPLS-DA model. Variables with a VIP value greater than 1 were further subjected to the nonparametric test (Wilcoxon, Mann–Whitney test). Only variables with VIP > 1.0 and p < 0.05 were selected as potential chemical markers. Structure annotation of the potential chemical markers was achieved by comparing the retention time and mass spectrum with those of the reference standards or by deducing information from their retention time and characteristic fragmentation patterns. A heat map was used to visualize the variation in the levels of the potential chemical markers in all of the DD and DA extracts.

2.7 Cytotoxic activities

In order to determine the cytotoxic efficacies of the chemical markers and the methanolic extracts, three human cancer cell lines, A549 (lung cancer cells), MCF7 (breast cancer cells), LoVo (colon cancer cells), and one mouse melanoma cells (B16) were undertaken. These cell lines were provided by the Institute of Botany, Jiangsu Province and Chinese Academy of Science, Nanjing, China, and were maintained in a humidified incubator at 37 °C in a 5% CO₂ atmosphere. The proliferation rates of the MCF7, A549, LoVo and B16 cells after treatment with the chemical markers and the methanolic extracts were determined using the colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay.¹⁸ The cells were plated at 9 × 10³ mL⁻¹ in 96-well plates with growth medium and incubated for 24 h at 37 °C with CO₂ in a humidified atmosphere. Various sample concentrations (10 μL) were added to the cultures after the transplantation. To determine the cell viability, 10 μL MTT (5 mg mL⁻¹) was added to each well, and the cells were cultured in additional incubation for 4 h. Then, the growth medium and the MTT reagent were removed, and DMSO (100 μL) was added to each well and then was shaken. cis-Diamminedichloroplatinum (CDDP, Sigma-Aldrich, USA) was used as positive control. The optical density was measured using an enzyme-linked immunosorbent assay (ELISA) reader at 570 nm. All the tests were performed in triplicate.

2.8 Antimicrobial activities

The antimicrobial activities of the chemical markers and the methanolic extracts were tested using three bacterial strains (Escherichia coli ATCC 8739, Staphylococcus aureus ATCC 6538 and Pseudomonas aeruginosa ATCC 16404), and two Fungal strains (Aspergillus niger ATCC 16404 and Monilia albican ATCC 10231). These microorganism strains were conserved at the Institute of Botany, Jiangsu Province and Chinese Academy of Science, Nanjing, China. The agar disc diffusion method was employed for the determination of antimicrobial screening¹⁹ with some modifications. A 100 μL suspension of the diluted bacterial strains (10⁶ CFU mL⁻¹) was spread on the solid media plates. The filter paper discs were impregnated with 10 μL samples and then placed onto the inoculated plates. After incubation for 24 h at 37 °C, all the plates were examined for any zones of growth inhibition, according to a method developed by Davidson and Parish (1989).²⁰ Cetylpyridinium chloride (CPC, Dr Ehrenstorfer GmbH, GER) served as a positive control. The minimum inhibitory concentrations (MIC) were determined using the broth microdilution method²¹ after 24 h for the microorganisms. The MIC was defined as the lowest concentration of the sample that completely inhibited the microbial growth.²² All tests were performed in triplicate.

2.9 Antioxidant activities

2.9.1 DPPH radical scavenging activity. The radical scavenging activities of the chemical markers and methanolic extracts of DD and DA were determined using 2,2-diphenyl-1-picrylhydrazyl (DPPH) as a reagent, according to the method of Burits and Bucar (2000)²³ with slight modifications. Briefly, 20 μL of various dilutions of the samples were mixed with 100 μL of a 0.16 mM ethanol solution of DPPH. After a 30 min incubation at room temperature, the scavenging capacity was determined spectrophotometrically by monitoring the decrease in the absorbance at 517 nm against a blank using a spectrophotometer (Shimadzu UV-1601, Kyoto, Japan). Vitamin C (Vc, Sigma-Aldrich, USA) was used as the positive control. All tests were performed in triplicate.

2.9.2 Ferric reducing antioxidant power (FRAP). The FRAP assay was performed as previously described by Oyaizu (1986).²⁴ In brief, various amounts (2.5, 5, 10, 20 and 40 μL) of the samples dissolved in 1.5 mL DMSO were mixed with 50 μL of 200 mM sodium phosphate buffer (pH 6.6) and 50 μL of 1% potassium ferricyanide. After being incubated at 50 °C for 20 min, 50 μL of 10% trichloroacetic acid was added to the mixture. The supernatant (100 μL) was mixed with distilled water (80 μL) and ferric chloride (20 μL, 1%). Finally, the absorbance was measured at 700 nm against a blank. Vc was used as the positive control and all tests were carried out in triplicate.

2.10 Statistical analysis for biological comparison

All analyses were carried out in triplicates and the data were expressed as the mean ± standard deviation (SD). Statistical differences were assessed from the one-way analysis of variance (ANOVA) and two-tailed Student's t-test. Values of p < 0.05 were considered statistically significant.

3. Results and discussion

3.1 UFLC-MS/MS analysis

Simultaneous determination of different types of compounds contained in DD and DA poses great challenge in separation and analysis time. Therefore, UFLC was used in our study because smaller column particle fillers and higher operation pressures could offer higher resolution and faster separation. Different types of mobile phases, such as acetonitrile–water system and methanol–water system with various modifiers, were compared. Methanol-0.1% aqueous formic acid was chosen as the best combination from the aspect of separation of major components and MS compatibility. For the mass detection, the positive and negative ion modes were compared to achieve maximum signals, and the results showed that higher sensitivity and more structural information could be obtained in the positive ion mode. The representative total ion chromatograms of the DD and DA extracts are presented in Fig. S1.†

In order to examine the system stability and reproducibility of the established method, the QC sample was analyzed for 5 times at the beginning of the run to ensure system equilibration for every 5 test samples to monitor system stability. Furthermore, the data quality was tested using the RSDs of peak intensity, retention time and m/z for all of the potential chemical markers. The RSD of the peak areas and the m/z of all of the potential chemical markers were less than 8% and 0.01%, respectively, and their retention times remained precisely same, which suggested that the established method was robust with good repeatability and stability, and could be used for chemical profiling of DD and DA.

3.2 Data pretreatment

Data pretreatment aims to eliminate the variance and bias introduced during the analysis of samples using mass spectrometric methods, thus extracting the most meaningful information from the raw data.²⁵ After the peaks had been picked with the MZ Mine 2 software and screened with “80% rule in test sample” and “RSD ≤ 30% in QC samples”, the data matrix corresponded to the total area that was normalized and Pareto-scaled before the chemometric analysis. Total area normalization could remove the unwanted systematic bias in the ion intensity within each sample during the measurements.^26,27 Pareto scaling was used to reduce the importance of the variables with large intensities as mass responses and the contents of the secondary metabolites in herbal medicines can range over many orders of magnitude, and compounds with higher contents or mass responses may exhibit high variation, exerting a greater influence on the multivariate projection models.^28,29

3.3 Multivariate analysis

The multivariate analysis techniques used in this study included PCA and OPLS-DA. PCA is a well-known unsupervised pattern-recognition method and is often used as the first step in chemometric analysis to visualize grouping trends and outliers. The variables pretreated as described above are transformed into new and uncorrelated variables called principal components (PCs).³⁰ A PCA plot for the first two PCs is usually used as these PCs capture most of the variation of the data. The closer the values of the PCs, greater is the similarity between the samples. As is shown in Fig. 2A, the very tightly clustered QC samples in the PCA scores plot indicated that the LC/MS system is stable throughout the entire analytical run. Clear separation could be observed between the DD and DA groups, and the highest separation was seen on the first PC with an explained variance of 66.6%.

Fig. 2 Scores plot of PCA (A) and OPLS-DA (B) based on data after pretreatment with statistical parameters as follows: (A) R2X = 0.798, Q2 = 0.652; (B) R2X = 0.757, R2Y = 0.994, Q2 = 0.989.

OPLS-DA is a supervised pattern-recognition method, which could be used to assist with the screening of potential chemical markers that account for such discrimination. The OPLS-DA model was validated using the seven-round internal cross-validation, and the model quality was described by the goodness-of-fit parameter R2, which represents how well the data fit the derived model, and the predictive ability parameter Q2.³¹ The model statistics of R2X, R2Y and Q2 are 0.757, 0.994, and 0.989, respectively, indicating that the models are robust and not the results of statistical over-fitting. The scatter plot is a window in the X space, in which the separation of the two classes of observations occurs in the horizontal direction (Fig. 2B). Hence, it is easier to uncover the variables with a class discriminating ability. Variables with VIP values greater than 1 were further subjected to the nonparametric test (Wilcoxon, Mann–Whitney test) to determine the significance of each variable. The vertical direction expresses within class variability, which is unrelated to the question (i.e., for discriminating between the two classes), but which is important for the complete understanding of the problem (Fig. 2B). Variables with VIP > 1.0 and p < 0.05 were selected as potential chemical markers, which were putatively confirmed by comparing the retention time and mass spectrum with those of 11 reference standards or by deducing from their retention time and characteristic fragmentation patterns. Detailed information of each chemical marker is shown in Table 1.

Table 1 Mass spectra properties of the chemical markers and their tentative identification results

No.	tR (min)	Adduct types	Adduct ions (m/z)	Main fragment ions (m/z)	Assigned identity	Formula	VIP value
a Confirmed with reference standards.
1^a	10.39	[M + Na]^+	301.35	—	Isodictamdiol	C15H18O5	1.06
2^a	14.64	[M + Na]^+	301.35	—	Dictamdiol	C15H18O5	1.48
3^a	19.63	[M + H]^+	233.15	215, 187, 123, 95	Fraxinellone	C14H16O3	1.93
4	20.05	[M + H]^+	261.15	243, 225, 215, 187, 119, 95	Calodendrolide	C15H16O4	3.13
5	15.75	[M + H]^+	487.25	469, 441, 383, 161, 95	Rutaevin	C26H30O9	1.78
6^a	16.05	[M + H]^+	471.25	453, 425, 161, 95	Limonin	C26H30O8	2.94
7^a	19.73	[M + H]^+	455.30	437, 409, 161, 95	Obacunone	C26H30O7	3.53
8^a	11.52	[M + H]^+	200.35	185, 128, 77	Isodictamnine	C12H9NO2	3.10
9	13.57	[M + H]^+	230.15	215, 200, 187	5 or 6 or 7-Methoxydictamnine	C13H11NO2	2.80
10^a	14.62	[M + H]^+	246.15	231, 216, 188	5-Hydroxy-4,8-dimethoxy furoquinoline	C13H11NO4	1.87
11	15.36	[M + H]^+	230.15	215, 200, 187	5 or 6 or 7-Methoxydictamnine	C13H11NO3	2.70
12^a	16.31	[M + H]^+	216.35	201, 145, 117, 90, 59	Robustine	C12H9NO3	2.50
13	16.72	[M + H]^+	260.15	245, 227, 216	Kokusaginin or maculosidin	C14H13NO3	1.40
14^a	17.18	[M + H]^+	260.15	245, 216	Skimmianine	C14H13NO4	2.40
15^a	17.43	[M + H]^+	230.15	215, 200, 187	γ-Fagarine	C13H11NO4	2.60
16^a	17.63	[M + H]^+	200.35	185, 129, 102	Dictamnine	C12H9NO2	2.20

---

3.4 Comparison of the relative concentration trends of the confirmed chemical markers

As shown in Fig. 3, these chemical markers could be subdivided into degraded limonoids (compound no. 1–3), limonoids (compound no. 4–6) and furoquinoline alkaloids (compound no. 7–16), according to their chemical structures. To illustrate the intensity trends of potential chemical markers, heat maps were employed to visualize the changes in the concentration in all samples (Fig. 4). In the heat map, the shade of colors indicates different levels of a chemical: the redder or greener the color is, the higher or lower the levels are, respectively. A value above zero represents higher concentration than the mean concentration in all samples and a value below zero represents a corresponding lower concentration. Most DD samples contain considerably higher levels of isodictamdiol, dictamdiol, fraxinellone, rutaevin, limonin, obacunone and dictamnine than the DA samples, whereas most DA samples contain considerably higher levels of isodictamnine, 5-hydroxy-4,8-dimethoxy furoquinoline, rubustine, skimmianine and γ-fagarine than the DD samples.

Fig. 3 Structural annotation of all the chemical markers, which could be subdivided into three chemical classes: degraded limonoids (compound no. 1–3), limonoids (compound no. 4–6) and furoquinoline alkaloids (compound no. 7–16).

Fig. 4 Heat map for the ion intensity trends of each chemical marker in all the samples. Shade of color indicates the concentration level of a chemical: the redder or greener the color is, the higher or lower is the concentration level, respectively. Hierarchical clustering analysis (HCA) was also applied to determine the similarities and differences among the 20 samples based on these chemical markers. DD = Dictamnus dasycarpus, DA = Dictamnus angustifolius.

In order to assess the discriminative ability of the identified chemical markers, PCA analysis based on 16 identified chemical markers was performed. After normalization and scaling, a 20 (observations) × 16 (variables) dataset was subjected to PCA analysis. The PCA bi-plot is shown in Fig. 5, in which both the observations and variables of the multivariate data are present in the same plot. Samples situated near the chemical markers are high in these markers, but are low in markers situated opposite. As depicted in Fig. 2A, all samples were successfully separated according to the species, based on the differences in the concentration levels of these chemical markers. The chemical composition of herbal medicines may vary with many factors such as the type of botanical species, growth locations, harvesting times and processing methods.³² DD distributes in many areas of China, whereas DA is only found in the Xinjiang province of China.^2,3 Therefore, the chemical difference between DD and DA could be attributed to the type of botanical species and to growing locations.

Fig. 5 PCA analysis based on 16 identified chemical markers was performed to assess the discriminative ability of the identified chemical markers. (A) Scores plot of PCA based on a 20 (samples) × 16 (chemical markers) dataset. (B) The bi-plot shows the correlation between the samples (yellow boxes and blue triangles) and the chemical markers (green dots). The closer the samples are to the chemical markers, the stronger is the correlation. Samples situated near the chemical markers are high in these markers and are low in markers situated opposite.

3.5 Cytotoxic activities

The cytotoxic activities of chemical markers and methanolic extracts of DD and DA were examined using the four mammalian cancer cell lines mentioned above. As shown in Table 2, the extracts of DA showed significant cytotoxic activity against in vitro MCF7 and LoVo cells with IC₅₀ values of 15.3 μg mL⁻¹ and 20.2 μg mL⁻¹, respectively. Although DD has been recommended for use in cancer therapy, its cytotoxicity against MCF7 (IC₅₀ = 28.2 μg mL⁻¹) and LoVo (IC₅₀ = 38.7 μg mL⁻¹) cells is relatively lower as compared to that of DA. The A549 and B16 cell lines both exhibited the lowest sensitivity to the extracts of DD and DA with IC₅₀ > l00 μg mL⁻¹.

Table 2 Cytotoxic activities of the chemical markers and methanolic extracts of DD and DA by MTT assay

Samples	Cell lines (IC50^a)
	A549	MCF7	LoVo	B16
a Values given as μg mL^−1.
1	>100	39.4 ± 2.0	45.2 ± 3.5	>100
2	>100	44.9 ± 1.6	48.1 ± 4.1	>100
3	55.1 ± 3.7	38.3 ± 2.3	35.4 ± 1.2	56.4 ± 2.5
6	>100	32.6 ± 3.4	37.3 ± 2.9	>100
7	>100	35.7 ± 1.8	41.7 ± 2.8	>100
8	54.7 ± 2.8	14.3 ± 1.4	18.6 ± 3.6	48.3 ± 2.2
10	>100	35.4 ± 1.2	32.8 ± 3.0	>100
12	>100	17.9 ± 2.0	24.3 ± 0.8	>100
14	>100	25.6 ± 3.2	33.1 ± 2.7	>100
15	>100	21.0 ± 2.6	29.4 ± 1.7	>100
16	>100	25.6 ± 3.2	27.9 ± 1.4	>100
DD	>100	28.2 ± 4.3	38.7 ± 2.9	>100
DA	>100	15.3 ± 1.1	20.2 ± 1.8	>100
CDDP	15.1 ± 0.9	9.5 ± 0.6	9.1 ± 1.3	14.4 ± 1.2

---

Statistically speaking, results from pharmacological studies (Table 2) suggested that furoquinoline alkaloids displayed more significant cytotoxic activities against MCF7 and LoVo cells than limonoids. However, fraxinellone (3) showed almost equally good activity as that of γ-fagarine (15). Previous research revealed that limonoids displayed significant cytotoxic activities against numerous mammalian cancer cell lines such as breast cancer cells,³³ carcinogen induced colon cancer cells,³⁴ neuroblastoma and colon adenocarcinoma cells.³⁵ Furthermore, there was also considerable research focused on the cytotoxic properties of furoquinoline alkaloids, including those against lung lymphoma L1210, epithelioid cervix carcinoma HeLa and other more extensive cell lines. Preskimmianine exhibited the most potent inhibitory activity with an IC₅₀ value of 3.1 μg mL⁻¹ in lung lymphoma.³⁶ In particular, another study has demonstrated that limonoids and furoquinoline alkaloids were also found to enhance the cytotoxicities of antineoplastic drugs such as vincristine and taxinol against the related tumor cells.³⁷ According to the comparative analysis of their chemical profiles, DA has a relatively higher level of many major furoquinoline alkaloids, whereas DD contains more degraded limonoids and limonoids. Therefore, the chemical difference between DD and DA may lead to their different cytotoxic activities.

3.6 Antimicrobial activities

The antimicrobial activities of chemical markers and methanolic extracts of DD and DA were evaluated against three bacterial strains and two fungal strains. As shown in Table 3, the extracts of DA showed a significant antimicrobial activity against the two bacterial strains E. coli ATCC 8739 and S. aureus ATCC 6538 with MIC values of 119 μg mL⁻¹ and 81 μg mL⁻¹ as well as against the two fungal strains A. niger ATCC 16404 and M. albican ATCC 10231 with MIC values of 263 μg mL⁻¹ and 56 μg mL⁻¹, respectively. However, as compared to DA, the antimicrobial activities of DD were relatively weak with MIC values of 178 μg mL⁻¹ and 152 μg mL⁻¹ for the two bacterial strains and 327 μg mL⁻¹ and 98 μg mL⁻¹ for the two fungal strains, respectively. In contrast, P. aeruginosa ATCC 16404 exhibited a minimal sensitivity to both DD and DA with MICs > 1000 μg mL⁻¹.

Table 3 Antimicrobial activities of the chemical markers and methanolic extracts of DD and DA by the agar disc diffusion and MIC methods

Samples	Microorganisms (MIC^a)
	Bacterial strains			Fungal strains
	Escherichia coli ATCC 8739	Staphylococcus aureus ATCC 6538	Pseudomonas aeruginosa ATCC 16404	Aspergillus niger ATCC 16404	Monilia albican ATCC 10231
a Values given as μg mL^−1.
1	219	183	>1000	306	115
2	203	197	>1000	372	128
3	192	151	923	249	91
6	271	160	>1000	296	147
7	246	179	>1000	337	183
8	76	44	663	163	25
10	187	134	>1000	222	109
12	110	92	>1000	209	63
14	144	103	>1000	285	84
15	128	79	>1000	218	79
16	109	86	847	272	87
DD	178	152	>1000	327	98
DA	119	81	>1000	263	56
CPC	15.8	<7.8	1000	204	<7.8

---

Because the extracts of DD and DA exhibit high antimicrobial activities, it is understandable that they have been used for the treatment of skin diseases such as eczema, rubella and scabies. In previous studies, limonoids such as fraxinellone have attracted considerable attention because of their various and significant biological properties, including antimicrobial activity.³⁸ However, compared with limonoid derivatives, furoquinoline alkaloids exhibited stronger antimicrobial activities against several human pathogenic bacteria and fungi.³⁹ From Table 3, it is clearly observed that the furoquinoline alkaloids exhibit more significant antimicrobial activities against E. coli ATCC 8739, S. aureus ATCC 6538, A. niger ATCC 16404 and M. albican ATCC 10231 than limonoids. The results mentioned above suggested that the difference in the antimicrobial activities between the furoquinoline alkaloids and the limonoid derivatives may result from their different chemical structures. It seems that the lactone groups in the limonoid derivatives are not critical for the growth inhibitory effect, whereas the furan ring and the surrounding oxygen-containing groups in the furoquinoline alkaloids might be important for such activity.⁷ The inhibitory effects of DA, quite understandably, were stronger in this experiment, because the extract of DA has a relatively higher concentration of many major furoquinoline alkaloids than that of DD.

3.7 Antioxidant activities

In the present study, the antioxidant activities of the chemical markers and methanolic extracts of DD and DA were determined by DPPH and FRAP assay. The scavenging ability of the two samples showed a concentration-dependent activity profile (Table 4). The extracts of DA showed marked antioxidant activity in the DPPH assay (78.3%) as compared to Vc (96.7%), whereas the extracts of DD (59.9%) showed moderate radical scavenging activities under this condition. The reducing power of the samples is presented in Table 5. As can be seen from the table, the extract of DA (0.9432 and 1.0214) and Vc (1.0288 and 1.0592) showed greater reducing power than that of DD (0.6309 and 0.8759) in 0.5 mg mL⁻¹ and 1.0 mg mL⁻¹, respectively.

Table 4 Scavenging abilities (%) of the chemical markers and methanolic extracts of DD and DA at different concentrations, determined by DPPH assay

Samples	Concentration^a (mg mL^−1)
	0.0625	0.125	0.25	0.5	1.0
a Values expressed are means ± S.D of three parallel measurements.
1	27.8 ± 1.2	38.5 ± 0.9	44.2 ± 1.2	66.4 ± 0.9	78.9 ± 1.7
2	21.5 ± 0.8	40.3 ± 1.6	52.5 ± 0.8	69.1 ± 1.4	79.8 ± 2.4
3	24.0 ± 1.3	31.6 ± 0.7	43.3 ± 1.6	65.9 ± 1.1	71.7 ± 2.1
6	23.2 ± 0.4	29.7 ± 1.1	39.1 ± 0.9	46.7 ± 1.5	54.3 ± 1.8
7	31.8 ± 0.5	36.2 ± 0.8	41.8 ± 2.3	52.5 ± 1.6	68.2 ± 2.0
8	39.1 ± 0.6	40.5 ± 1.4	53.7 ± 0.9	67.4 ± 1.9	71.2 ± 2.6
10	47.3 ± 1.7	45.6 ± 2.6	63.5 ± 1.4	79.1 ± 2.7	82.8 ± 3.5
12	48.7 ± 2.2	51.4 ± 1.3	69.7 ± 1.6	83.2 ± 3.0	87.3 ± 1.7
14	48.2 ± 1.9	41.7 ± 0.8	52.8 ± 2.8	64.7 ± 1.3	73.5 ± 1.4
15	46.4 ± 0.9	48.3 ± 1.3	57.2 ± 2.1	68.6 ± 0.9	79.7 ± 1.8
16	43.1 ± 1.4	49.5 ± 1.9	53.7 ± 0.6	65.9 ± 2.3	68.2 ± 0.9
DD	27.3 ± 1.6	30.6 ± 2.0	42.3 ± 1.1	49.7 ± 0.7	59.9 ± 1.2
DA	44.5 ± 0.8	41.4 ± 0.6	58.1 ± 2.5	71.6 ± 1.4	78.3 ± 2.2
Vc	54.0 ± 1.7	83.5 ± 0.4	93.3 ± 0.8	95.8 ± 1.2	96.7 ± 2.3

---

Table 5 Reducing power of the chemical markers and methanolic extracts of DD and DA at different concentrations

Samples	Concentration^a (mg mL^−1)
	0.0625	0.125	0.25	0.5	1.0
a Values expressed are means ± S.D of three parallel measurements.
1	0.4142 ± 0.021	0.4317 ± 0.013	0.6039 ± 0.008	0.7391 ± 0.011	0.9559 ± 0.025
2	0.4027 ± 0.016	0.4983 ± 0.021	0.5991 ± 0.019	0.6979 ± 0.024	1.0141 ± 0.018
3	0.3236 ± 0.014	0.4161 ± 0.008	0.5362 ± 0.013	0.5882 ± 0.029	0.7268 ± 0.016
6	0.4073 ± 0.027	0.4752 ± 0.012	0.6095 ± 0.021	0.6279 ± 0.015	0.8957 ± 0.023
7	0.3385 ± 0.012	0.3819 ± 0.023	0.5913 ± 0.010	0.6242 ± 0.012	0.7891 ± 0.017
8	0.3938 ± 0.009	0.4023 ± 0.013	0.5569 ± 0.021	0.8997 ± 0.014	1.0127 ± 0.015
10	0.4387 ± 0.018	0.4842 ± 0.032	0.6479 ± 0.021	0.6243 ± 0.018	0.9599 ± 0.020
12	0.4465 ± 0.012	0.4927 ± 0.027	0.6849 ± 0.023	0.9587 ± 0.016	1.0102 ± 0.022
14	0.4036 ± 0.024	0.5172 ± 0.019	0.6565 ± 0.019	0.9279 ± 0.007	1.0065 ± 0.019
15	0.4224 ± 0.018	0.4326 ± 0.017	0.6824 ± 0.033	0.9142 ± 0.012	0.9891 ± 0.023
16	0.4212 ± 0.027	0.4473 ± 0.014	0.5895 ± 0.018	0.8247 ± 0.009	0.9729 ± 0.017
DD	0.4027 ± 0.016	0.4524 ± 0.022	0.6167 ± 0.028	0.6309 ± 0.021	0.8759 ± 0.025
DA	0.4316 ± 0.011	0.3971 ± 0.026	0.5731 ± 0.013	0.9432 ± 0.018	1.0214 ± 0.014
Vc	0.8063 ± 0.034	0.9417 ± 0.009	0.9546 ± 0.011	1.0288 ± 0.004	1.0592 ± 0.010

---

Antioxidant research (Tables 4 and 5) on chemical markers and methanolic extracts present in the two plants advocated that furoquinoline alkaloids possessed substantial antioxidant activity, whereas limonoids exhibited a relatively weak activity. Previous studies found that the antioxidant activity could be attributed to the total phenolic content in the selected herbs.^40,41 Because furoquinoline alkaloids contain one or more aromatic rings bearing one or more hydroxy groups, they also belong to phenolic compounds structurally. According to the chemometric analysis of the two medicinal herbs, DA has a higher level of furoquinoline alkaloids, whereas DD contains more limonoids. Therefore, the higher antioxidant activity of DA than DD may possibly be attributed to the presence of a high level of furoquinoline alkaloids in it.

4. Conclusion

In this report, we first investigated the chemical difference between DD and DA using UFLC-MS/MS-based plant metabolomics approaches, and then compared their cytotoxic, antimicrobial and antioxidant activities, which are closely related to their therapeutic effects in clinics. A total of 16 chemical markers (4 degraded limonoids, 3 limonoids and 9 furoquinoline alkaloids) responsible for species differentiation were screened out. The relative concentration trends of each chemical marker indicated that there were significant differences in the chemical composition between these two species, which may account well for their different cytotoxic, antimicrobial and antioxidant activities. Among all the chemical markers, furoquinoline alkaloids (particularly isodictamnine), exhibiting stronger activities, may play a major role in the bioactivities of these two plants. The findings might not only offer very useful information for the authentication and quality control of these two species, but also will lay a foundation for the safe and efficacious use of crude extracts from DD and DA in clinics.

Acknowledgements

This study was financially supported by the National Science Foundation of China (no. 81302733), the research project of Chinese Ministry of education (no. 113036A), the program for Jiangsu province Innovative Research Team, the Program for New Century Excellent Talents in University (no. NCET-13-1036), the Fundamental Research Funds for the Central Universities (no. JKZD2013004) and the Open Project Program of State Key Laboratory of Natural Medicines, China Pharmaceutical University (no. SKLNMZZYQ201303, SKLNMKF201220).

Notes and references

1. J. B. Sun, W. Qu, Y. Xiong and J. Y. Liang, Biochem. Syst. Ecol., 2013, 50, 62–64.
2. J. L. Yang, L. L. Liu and Y. P. Shi, Planta Med., 2011, 77, 271–276.
3. T. S. Wu, C. Y. Li, Y. L. Leu and C. Q. Hu, Phytochemistry, 1999, 50, 509–512.
4. V. I. Akhmedzhanova, I. A. Bessonova and S. Y. Yunusov, Chem. Nat. Compd., 1978, 14, 404–406.
5. J. B. Sun, W. Qu, P. Wang, F. H. Wu, L. Y. Wang and J. Y. Liang, Fitoterapia, 2013, 90, 209–213.
6. J. S. Yoon, S. H. Sung and Y. C. Kim, J. Nat. Prod., 2008, 71, 208–211.
7. W. Zhao, J. L. Wolfender, K. Hostettmann, R. Xu and G. Qin, Phytochemistry, 1998, 47, 7–11.
8. J. Kim, G. K. Jayaprakasha and B. S. Patil, Biochimie, 2014, 105, 36–44.
9. L. W. Qi, H. Y. Wang, H. Zhang, C. Z. Wang, P. Li and C. S. Yuan, J. Chromatogr. A, 2012, 1230, 93–99.
10. L. Geng, H. Sun, Y. Yuan, Z. Liu, Y. Cui, K. Bi and X. Chen, Fitoterapia, 2013, 84, 286–294.
11. A. C. Martin, A. D. Pawlus, E. M. Jewett, D. L. Wyse, C. K. Angerhofer and A. D. Hegeman, RSC Adv., 2014, 4, 26325–26334.
12. P. H. Zhao, L. M. Sun, M. A. Cao and C. S. Yuan, Chem. Lett., 2007, 36, 1200–1201.
13. Z. L. Liu, Y. J. Xu, J. Wu, S. H. Goh and S. H. Ho, J. Agric. Food Chem., 2002, 50, 1447–1450.
14. J. Tang, W. Zhu and Z. Tu, Zhongcaoyao, 1995, 26, 563–565.
15. T. Pluskal, S. Castillo, A. Villar-Briones and M. Oresic, BMC Bioinf., 2010, 11, 395.
16. S. Bijlsma, I. Bobeldijk, E. R. Verheij, R. Ramaker, S. Kochhar, I. A. Macdonald, B. van Ommen and A. K. Smilde, Anal. Chem., 2006, 78, 567–574.
17. M. Wang, Y. Li, Y. Huang, Y. Tian, F. Xu and Z. Zhang, J. Sep. Sci., 2014, 37, 1148–1154.
18. T. Mosmann, J. Immunol. Methods, 1983, 65, 55–63.
19. A. Ferrero, A. Minetti, C. Bras and N. Zanetti, J. Ethnopharmacol., 2007, 113, 441–447.
20. P. Davidson and M. Parish, Food Tech., 1989, 43, 148–155.
21. NCCLS, Performance Standards for Antimicrobial Susceptibility Testing: Twelfth Informational Supplement, document M100-S12, Wayne, Pennsylvania 19087–21898 USA, 2003.
22. A. Smânia, F. D. Monache, E. Smânia, M. Gil, L. Benchetrit and F. Cruz, J. Ethnopharmacol., 1995, 45, 177–181.
23. M. Burits and F. Bucar, Phytother. Res., 2000, 14, 323–328.
24. M. Oyaizu, Jpn. J. Nutr., 1986, 44, 307–315.
25. M. Ernst, D. B. Silva, R. R. Silva, R. Z. Vencio and N. P. Lopes, Nat. Prod. Rep., 2014, 31, 784–806.
26. Y. Li, M. Wang, Q. Zhang, Y. Tian, F. Xu and Z. Zhang, Anal. Methods, 2014, 6, 1720–1727.
27. X. Zhao, F. Xu, B. Qi, S. Hao, Y. Li, Y. Li, L. Zou, C. Lu, G. Xu and L. Hou, J. Proteome Res., 2014, 13, 1101–1111.
28. C. Ducruix, D. Vailhen, E. Werner, J. B. Fievet, J. Bourguignon, J. C. Tabet, E. Ezan and C. Junot, Chemom. Intell. Lab. Syst., 2008, 91, 67–77.
29. Y. Huang, Y. Tian, G. Li, Y. Li, X. Yin, C. Peng, F. Xu and Z. Zhang, Anal. Bioanal. Chem., 2013, 405, 4811–4822.
30. M. Rafi, L. W. Lim, T. Takeuchi and L. K. Darusman, Anal. Bioanal. Chem., 2013, 405, 6599–6603.
31. W. B. Dunn, D. Broadhurst, P. Begley, E. Zelena, S. Francis-McIntyre, N. Anderson, M. Brown, J. D. Knowles, A. Halsall, J. N. Haselden, A. W. Nicholls, I. D. Wilson, D. B. Kell, R. Goodacre and C. Human Serum Metabolome, Nat. Protoc., 2011, 6, 1060–1083.
32. J. Deng and Y. Yang, Anal. Chim. Acta, 2013, 785, 82–90.
33. Q. Tian, E. G. Miller, H. Ahmad, L. Tang and B. S. Patil, Nutr. Cancer, 2001, 40, 180–184.
34. T. Tanaka, M. Maeda, H. Kohno, M. Murakami, S. Kagami, M. Miyake and K. Wada, Carcinogenesis, 2001, 22, 193–198.
35. S. M. Poulose, E. D. Harris and B. S. Patil, Nutr. Cancer, 2006, 56, 103–112.
36. X. Gao, P. H. Zhao and J. F. Hu, Chem. Biodiversity, 2011, 8, 1234–1244.
37. A. Roy and S. Saraf, Biol. Pharm. Bull., 2006, 29, 191–201.
38. C. Yuan, X. Wang and D. Yang, Mortality, 2006, 24, 72h.
39. B. Wolters and U. Eilert, Planta Med., 1981, 43, 166–174.
40. Y. Velioglu, G. Mazza, L. Gao and B. Oomah, J. Agric. Food Chem., 1998, 46, 4113–4117.
41. Y. Cai, Q. Luo, M. Sun and H. Corke, Life Sci., 2004, 74, 2157–2184.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra00115c

---